Benjamin Bossan

Okay, so I investigated the failing tests with PyTorch 2.4 a bit further. Right now, we get a warning: > FutureWarning: You are using `torch.load` with `weights_only=False` (the current default...

Thanks for the input, this sounds reasonable. It's not pretty, but since we cannot directly pass arguments to `__setstate__`, I don't see a better way. As to the default: WDYT...

Okay, so if I understand your question correctly, you have run a grid search and determined the best hyper-parameters. Now you would like to take those best parameters and train...

Just to be clear, grid search (and similar methods like randomized search) are intended to figure out the best hyper-parameters, which I think is what you intend to do here....

> My sample size is small, total of 616 , 437 for cross validation and 179 for test the final model. This is indeed very small and I assume you...

So I assume that just using `peft_model.save_pretrained` does not work for you? You could check the state dict returned by [`get_peft_model_state_dict`](https://huggingface.co/docs/peft/v0.15.0/en/developer_guides/low_level_api#saving-the-model), maybe write a function that adds the missing pieces...

Adding support for more layers would indeed be a nice addition to PEFT. If you already have an implementation for IA³, please feel free to open a PR. When it...

Resolved via #2082.

Maybe not relevant, but Python added its own [`warnings.deprecated`](https://docs.python.org/3/library/warnings.html#warnings.deprecated) decorator in Python 3.13. Using the same name could be confusing. The Python decorator apparently also allows type checkers to flag...

As to the Python version, 3.8 is going to be end of life by the end of the month, so the containers will be updated soon. Still, it will be...